Fudan MOSS team core members release a brand new multi-agent system that can automatically complete the entire research process from literature review to paper writing.

Scientific research activities have high barriers to entry and high trial and error costs. In the existing academic system, a large number of "failed" results are often not publicly recorded, making it impossible for different researchers to explore them repeatedly. Automated research systems based on large language models are expected to alleviate this problem. FARS, a multi-intelligent system developed by Shanghai Rixingji Intelligent Technology Co., Ltd., founded by core members of the MOSS team at Fudan University in March last year, was released on February 12th. This system is capable of fully automating the entire research process, including literature research, hypothesis generation, code writing, experiment execution, and paper writing, with the potential to make scientific discovery more efficient, scalable, and reduce research costs.